What's known on the subject? and What does the study add?
The use of MRI for prostate cancer diagnosis and staging is increasing. Indications for prostate MRI are not defined and many clinicians are unsure of how best to use MRI to aid clinical decisions.
This evidence-based medicine article addresses the clinical utility of prostate MRI for preoperative staging. Based on a common patient scenario, a guide to calculating the probability of extraprostatic extension is provided.
A 61-year-old healthy male presents to your clinic with a serum PSA concentration of 4.6 ng/mL. The DRE is abnormal with a firm left prostate lobe (cT2a). He undergoes a 10-core biopsy that reveals four positive cores from the left lobe (one core Gleason 3 + 3, three cores Gleason 3 + 4, all ≤10% cancer in each core), and one positive core from the right lobe (Gleason 3 + 3, 10% cancer). The patient thinks that radical prostatectomy is his best treatment option. His first priority is cancer cure, but he is also interested in preserving his urinary and sexual function. You explain that preservation of his periprostatic neurovascular structures will increase his chance of postoperative erectile function but may increase his risk of a positive surgical margin, especially if there is extraprostatic tumour extension (EPE). He specifically asks if pelvic MRI may help predict if he has EPE. You tell the patient that you usually do not obtain preoperative pelvic MRI and you are unfamiliar with how it changes his risk beyond what we know from his physical examination and biopsy information. You decide to review the relevant literature on the diagnostic performance of MRI for detecting EPE and commit to the patient that you will get back to him with a more informed opinion.
For patients with prostate cancer, how does prostate MRI impact the probability of EPE?
Finding the Best Evidence
You begin your search by reviewing an article from the Users' Guide to the Urological Literature titled ‘How to Use an Article about a Diagnostic Test’ . From this article, you realise that the first step is to determine the patient's risk of EPE based on his clinical and biopsy findings. Then you need to find the best evidence on the performance of MRI in patients with prostate cancer.
The pre-test probability is your level of suspicion that this patient has EPE. In this case, we do not have to rely on our ‘gut-feeling’ because there are nomograms that help predict risk of EPE based on available clinical, PSA and biopsy characteristics. Potentially even more useful in our situation, given that we can preserve or resect the neurovascular bundle on either side, is a nomogram that predicts the risk of EPE on each side of the prostate. According to the Partin tables, our patient has a 41% estimated risk of EPE . But using the side-specific nomogram we realise he has an 8% risk on the right and a 35% risk on the left .
Sensitivity and Specificity
To calculate the likelihood ratios (LRs) associated with prostate MRI you require information about the sensitivity and specificity of the test. You perform a PubMed search using the terms ‘MRI’ and ‘prostate cancer’ yielding >2900 articles. However, limiting your search to systematic reviews yields only seven articles. Closer review of your results reveal two articles investigating the accuracy of MRI imaging for prostate cancer staging which appear suitable to answer your question [4, 5].
The more recent and comprehensive systematic review included 146 studies examining the sensitivity and specificity of MRI for local staging of prostate cancer when compared with pathology, the reference standard. The systematic review showed a wide range of sensitivities and specificities between studies. The authors of the systematic review performed a meta-analysis of studies to calculate an estimate of the performance of MRI. This can be thought of as the ‘average’ performance among studies with a sensitivity of 71% and specificity of 71% . You discover that the staging performance of MRI is influenced by several factors including publication year, number of imaging planes, endorectal coil usage, and very importantly, the experience of the reading practitioner [4, 6]. As recognised by the authors, you conclude that this meta-analysis is limited due to the very heterogeneous nature, and occasional poor quality, of the studies. However, as this analysis included a large and diverse selection of patients (and is the best available), you plan to use the estimates from the review for your calculations.
LRs represent how a diagnostic test increases or decreases the probability that your patient has the disease of interest . In this example, LRs allow you to determine how MRI changes the probability that your patient has EPE. As you know the estimated sensitivity (0.71) and specificity (0.71) of MRI, you are able to calculate the positive (LR+) and negative (LR−) LRs (Appendix). The LR+ allows you to estimate the probability of EPE when MRI suggests tumour extension and the LR− allows you to estimate the probability of EPE when the MRI is normal.
Case Resolution – Applying the Results to the Care of Your Patient
You call the patient to discuss your findings. You preface your discussion by informing the patient that the literature evaluating the performance of prostate MRI has limitations and should be interpreted with caution. You inform him that the probability of EPE on the right is 8% and a positive MRI will only increase his risk to 17%. However, the probability on the left side is 35% and this risk will increase to 57% if the MRI is positive and will decrease to 18% if the MRI is negative. The patient thinks that he will choose wide resection of the left neurovascular bundle if his risk of tumour extension is 57% (positive MRI). On the other hand, if his risk is only 18% on the left he will request bilateral nerve-preservation. Based on this informed discussion about his risk and preferences, you agree to obtain a pelvic MRI.
The endorectal coil MRI suggests EPE on the left side (Fig. 1) . Using his pre-test probability of EPE (35%) and the LR+ (2.45), you confirm his post-test probability using a Fagan nomogram (Fig. 2) . The patient asks you to proceed with nerve-sparing on the right side and nerve resection on the left. Pathological analysis reveals a pT3a tumour (EPE on left) with negative surgical margins.
The above case shows the benefit and limitations of prostate MRI to predict EPE and outlines how to use diagnostic test performance concepts to aid clinical decision-making. Based on the current evidence, prostate MRI may be beneficial in select patients with intermediate probability of EPE. The benefit of MRI in patients with low risk is probably minimal because MRI lacks the ‘power’ to significantly change the probability of EPE. Most importantly, you are now aware of the limitations of currently available evidence assessing pelvic MRI and look forward to the results of future studies evaluating prostate cancer imaging.
Rodney H. Breau is supported by the Canadian Urological Association Scholarship Foundation, the AUA Northeastern Section Young Investigator Award, and Prostate Cancer Canada Network.
Conflict of Interest
Appendix: Appendix: Calculation of LRS for MRI and Prostate Cancer